The first rule of experiments is “do no harm.” Vouchers fail that test.

As we know from the voucher programs in Cleveland and Milwaukee, and voucherlike systems in Chile, Sweden and the Netherlands, vouchers siphon money from the public schools that the vast majority of children attend.

In Milwaukee, the drain on public schools is $29 million this year alone.

The second rule of experiments is “don’t do them when you already know the results.”

In the case of vouchers, all of the evidence shows that not only do they drain resources from public schools, they also don’t improve student achievement -- certainly not for poor children.

There’s another rule, a moral one: When you know what works in education, do it.

For example, we know without doubt that reducing class size in the early grades improves student achievement, especially for inner- city students, and these positive effects last throughout high school.

In Milwaukee, public school elementary students in small classes outperformed then counterparts in voucher schools.

We should encourage more schools to adopt programs proven to boost student learning, like the “Success for All” elementary reading program developed by Johns Hopkins University. More than a decade of research shows it works.

For the $10 million Cleveland spent in 1998 on vouchers for 3,000 Cleveland students, the city could put “Success for All” in every elementary school, boosting the achievement of 40,000 Cleveland youngsters - with $6 million left over.

Whether or not you call vouchers an “experiment,” it makes no sense to spend millions of public dollars to send a select handful of children to private schools when those same dollars can fund proven programs in the public schools that 90% of our children attend.

Sandra Feldman is president of the American Federation of Teachers.